A decorrelation preprocessing method based on linear predictive coding (LPC) is proposed for enhancing entropy-source signals from a Brillouin random fiber laser (BRFL). Unlike structure-optimization-based approaches, the proposed algorithm directly suppresses predictable components in raw BRFL entropy-source signals by extracting prediction residuals from historical-sample-based linear prediction. Experimental results show that LPC preprocessing significantly reduces the autocorrelation sidelobe ratio of the BRFL sampling sequence from −0.044 to −28.917dB, increases the spectral flatness from 0.005 to 0.952, and improves the amplitude min-entropy from 5.454 to 6.190 bit/sample. After random-bit postprocessing, the generated bitstream exhibits good 0/1 balance and low residual correlation and successfully passes the NIST randomness tests. These results demonstrate the effectiveness of LPC-based decorrelation preprocessing in enhancing BRFL entropy-source signals, indicating its potential for fiber-laser-based physical random number generation and secure information processing.
A decorrelation preprocessing method based on linear predictive coding (LPC) is proposed for enhancing entropy-source signals from a Brillouin random fiber laser (BRFL). Unlike structure-optimization-based approaches, the proposed algorithm directly suppresses predictable components in raw BRFL entropy-source signals by extracting prediction residuals from historical-sample-based linear prediction. Experimental results show that LPC preprocessing significantly reduces the autocorrelation sidelobe ratio of the BRFL sampling sequence from −0.044 to −28.917dB, increases the spectral flatness from 0.005 to 0.952, and improves the amplitude min-entropy from 5.454 to 6.190 bit/sample. After random-bit postprocessing, the generated bitstream exhibits good 0/1 balance and low residual correlation and successfully passes the NIST randomness tests. These results demonstrate the effectiveness of LPC-based decorrelation preprocessing in enhancing BRFL entropy-source signals, indicating its potential for fiber-laser-based physical random number generation and secure information processing.
High-speed physical random number generation is essential for secure communication, high-speed encryption and stochastic photonic information processing. However, scaling optical random number generators toward massively parallel operation remains challenging, because multiple wavelength channels generated from a common physical entropy source often suffer from residual correlation and limited spectral scalability. Here, we propose and experimentally demonstrate a dual-output, multi-channel parallel physical true random number generator based on a randomdistributed-feedback Brillouin fiber laser. Cascaded stimulated Brillouin scattering, four-wave mixing, cross-phase modulation and self-phase modulation in a highly nonlinear fiber, together with near-zero-dispersion assisted phase matching provided by a dispersion-shifted fiber, generate broadband Brillouin optical frequency combs spanning approximately 100 nm at both output ports. Distributed Rayleigh backscattering in a long single-mode fiber introduces stochastic phase perturbations and optical path delays, effectively breaking the fixed coherent relationship between output1 and output 2. The two outputs simultaneously exhibit broadbandcomb spectra, withoptical signal-to-noise ratio values of approximately 20 dB observed for the comb lines. The temporalcross-correlation coefficient between the two outputs remains below 0.05, forming two decorrelated optical entropy sources. Using 40 GSa/s sampling, 8-bit quantization, delay-reverse XOR decorrelation, and 4-LSB extraction, eight representative comb-line channels from the two output ports pass the NIST SP 800-22 statistical test suite. Based on the adopted offline post-processing procedure, these channels correspond to an effective aggregate random-bit rate of 1.92 Tb/s. The broadband spectra contain more than 1100 spectrally observable comb lines at each output port, indicating substantial potential for wavelength-parallel scaling. Under anideal fully parallel acquisition and processing architecture, the corresponding theoretical upper-bound aggregate rate could exceed 352 Tb/s (2200 × 4bit × 40GSa/s).
A bidirectional pumping Brillouin random fiber laser based on a non-uniform fiber is proposed and demonstrated for truly random number generation with enhanced bit rate. Numerical simulation indicates that, thanks to the unique distributed Brillouin gain profile along the non-uniform fiber, random lasing oscillation within a broadband Brillouin gain can be established by the mutual effect of Brillouin scattering and enhanced Rayleigh scattering, providing physical entropy source for truly random number generation. The poof-of-concept experiments demonstrate a truly random number with a bit rate of 71 Mbps, which significantly breaks through the limitation of the intrinsic Brillouin bandwidth (~10 MHz) in communication fibers.
True random numbers are essential for cryptography and secure communications. Algorithmically generated pseudo-random numbers lack true physical randomness, while conventional lasers as physical entropy sources introduce time-delay signatures (TDS) and require complex post processing. To address this challenge, this study presents a high speed, real-time true random bit generator utilizing a modeless Raman fiber laser (RFL). A modeless RFL was constructed using an ultra-low reflectivity (-27 dB) fiber Bragg grating (FBG) and a high-reflectivity FBG serving as cavity mirrors. The laser delivers stable modeless Raman lasing within the pump power range of 5 W to 10 W. Compared with multilongitudinal-mode operation, the modeless laser exhibits significantly lower temporal coherence. By truncating the four least significant bits of the 8-bit digitized temporal intensity signal from the modeless RFL, random bit sequences were generated at a rate of 320 Gbps. The resulting random bit stream shows a uniform distribution of bits 0 and 1, its autocorrelation function contains no TDS, and it successfully passes the NIST statistical test. In contrast, the same generation method fails to produce qualified random bits when applied to the Raman fiber laser operating in the multilongitudinal-mode regime. The demonstrated modeless RFL system combines real-time operation, high speed, and stability, offering a robust physical entropy source for cryptographic and secure communication applications.
True random numbers occupy a core role in the field of encryption and security by virtue of their properties such as unpredictability and independence. In this paper, we propose a high-speed physical random number generator based on a dual-emission random fiber laser. The system improves the spectral bandwidth characteristics of the output signals by adjusting the pump power and the feedback length of the single mode fiber. This study effectively suppresses the autocorrelation feature by performing exclusive-OR operations on the forward and backward random laser signals, and obtain the random number sequence with a generation rate of 20 Gbps by post-processing with high-order finite difference and least significant bit extraction. In order to verify the practical application performance, we apply the random sequence as encryption key to digital image encryption, and the experimental results show that the generator has both theoretical security and practical application value.
A simple novel method for random number generation is presented, based on a random Raman fiber laser. This laser is built in a half-open cavity scheme, closed on one side by a narrow-linewidth 100 mm fiber Bragg grating. The interaction between the randomly excited lasing modes of this laser, in addition to nonlinear effects such as modulation instability, allow the generation of random bits at rates of up to 540 Gbps with minimal post processing. Evaluation of the resulting bit streams' randomness by the NIST statistical test suite highlights the importance of evaluating the physical entropy content, as bit sequences generated by this random laser pass all the statistical tests with a significance level of 0.01, despite being generated at more than twice the theoretical entropy generation speed.
True random bits are crucial for cryptographic systems, particularly in key generation and secure initialization. In this paper, we present a novel technique based on chaotic signals and the addition of pulse-amplitude modulation signals to generate random bits. This technique can be implemented directly in physical systems. The randomness of our generated bit sequences will be examined with the National Institute of Standards and Technology (NIST) statistical test. The results will also be compared with a widely used technique reported in the literature.